Every N seconds, a fixed count, or every single frame. Drop, configure, click.
Drop your video into PerfectStudio's Frames mode, pick an extraction strategy (Interval, Count, or Every frame), pick output format (JPG / PNG / WebP), click Extract. Free tier handles this guide indefinitely — outputs carry a centred watermark you can remove with Pro ($59 one-time). Studio tier ($129) adds folder-level batching.
Launch the app. From the mode tabs at the top, pick Frames — the dedicated frame-extraction workspace, separate from Aspect / GIF / Slow-Mo.
Drag the source onto the drop zone, or click to browse. MP4, MOV, AVI, MKV, WebM all work. Studio tier adds ProRes and DNxHR input — useful when you're pulling stills from edit-grade source.
Interval: one frame every N seconds — e.g., 1.0 for once per second, 0.5 for twice per second. Best for tagging libraries and consistent thumbnails. Count: N frames evenly spaced across the whole clip — e.g., 12 for a Vimeo-style thumbnail strip. Every frame: every single frame at the source frame rate. Best for animation review or ML training data — be ready for the output size.
JPG: smallest, lossy, no alpha — best for thumbnails and tagging. PNG: lossless, supports alpha — best for clean stills and overlays. WebP: small like JPG, lossless option like PNG, alpha — modern default if your downstream tooling supports it.
Defaults to a subfolder next to the source. Sequential names — frame_001.jpg, frame_002.jpg, ... — keep order predictable. Pick a dedicated folder if you want to drop the result into a tagging tool, a static-site image directory, or a training dataset.
The bundled FFmpeg pipeline scans the video and writes each frame as it lands. Progress shows live. A 60-second video at 1 frame/sec finishes in about 5 seconds on Apple Silicon and produces 60 files.
frame_001.jpg at 1 fps interval = 0.0 seconds, frame_002.jpg = 1.0 seconds, and so on.JPG for thumbnails, tagging libraries, or ML training pipelines. PNG for clean stills, transparency, or downstream Photoshop work. WebP for the best of both — small like JPG, lossless option like PNG, alpha support.
Thumbnails: 5-12 evenly spaced. Animated preview: 12-30. ML training: every frame or every 2nd. Dialogue tagging: 1 per second catches most distinct shots. Count mode picks the exact target number; Interval picks the frequency.
Faster than expected. 30-second 4K at every-frame PNG is ~3 GB. 1-per-second JPG is under 1 MB. Pick the strategy that matches your use case.
Yes — separate signed DMG for each chip. Apple Silicon is faster on large extractions because the bundled FFmpeg uses VideoToolbox where the codec supports it. Intel works fine, just slower on 4K+ source.
Yes — Studio tier ($129) drops a folder, sets the config once, queues every video. Each video's output lands in its own subfolder.
Use Interval with a tight number (0.1 seconds) and pick the closest, or use Count + manual scrub. A dedicated 'pull frame at 00:01:23.456' control is on the roadmap.
Comparing tools instead? See the PerfectStudio comparison hub — vs ezgif, HandBrake, Twixtor, CloudConvert, Adobe Media Encoder, FFmpeg, Photoshop Batch, Topaz Video AI, DaVinci Resolve.
Aspect mode, GIF chunker, frame extractor, AI slow-motion — all unlocked. Watermark on outputs until you upgrade. Mac (Apple Silicon + Intel) and Windows.